減去前一行的數量MySQL

Question

因此而艱難。我在PHP中迴應了這些數據，但我認爲在MySQL中進行查詢會更好。

這裏是我的表：

我的查詢（錯誤）：

SELECT i1.Quantity, 
i1.timestamp, 
(i2.Quantity - i1.Quantity) as sold_qty 
FROM InventoryTest_history i1 
INNER JOIN 
InventoryTest_history i2 ON i2.id = i1.id + 1 
WHERE i1.SKU = '(L) U-Joint' 

僅檢索兩行，而sold_qty是錯誤的。

我想實現是這樣的結果：

Quantity | timestamp | sold_qty 
985   2016-12-27  0 
960   2016-12-28  25 
955   2016-12-29  5 

誰能幫助？

Answer 1

你也可以使用這個沒有JOIN表：出SELECT只是隱藏一些列。

SELECT Quantity, `timestamp`, sold_qty 
FROM (
    SELECT i.*, 
    @sold_qty := GREATEST(@last_qty - i.`Quantity`,0) as sold_qty, 
    @last_qty := i.`Quantity` as last_qty 
    FROM InventoryTest_history i 
    CROSS JOIN (SELECT @last_qty := '', @sold_qty := 0) as init 
    ORDER BY `timestamp` 
    ) as result; 

樣品

mysql> SELECT * from InventoryTest_history; 

+----+-------------+----------+---------------------+ 
| id | SKU   | Quantity | timestamp   | 
+----+-------------+----------+---------------------+ 
| 1 | (L) U-Joint |  985 | 2016-12-27 10:08:58 | 
| 2 | (L) U-Joint |  960 | 2016-12-28 10:09:52 | 
| 3 | (L) U-Joint |  955 | 2016-12-29 16:01:02 | 
+----+-------------+----------+---------------------+ 
3 rows in set (0,02 sec) 

mysql> SELECT Quantity, `timestamp`, sold_qty 
    -> FROM (
    ->  SELECT i.*, 
    ->  @sold_qty := GREATEST(@last_qty - i.`Quantity`,0) as sold_qty, 
    ->  @last_qty := i.`Quantity` as last_qty 
    ->  FROM InventoryTest_history i 
    ->  CROSS JOIN (SELECT @last_qty := '', @sold_qty := 0) as init 
    ->  ORDER BY `timestamp` 
    -> ) as result; 
+----------+---------------------+----------+ 
| Quantity | timestamp   | sold_qty | 
+----------+---------------------+----------+ 
|  985 | 2016-12-27 10:08:58 |  0 | 
|  960 | 2016-12-28 10:09:52 |  25 | 
|  955 | 2016-12-29 16:01:02 |  5 | 
+----------+---------------------+----------+ 
3 rows in set (0,00 sec) 

mysql> SELECT i.*, 
    -> @sold_qty := GREATEST(@last_qty - i.`Quantity`,0) as sold_qty, 
    -> @last_qty := i.`Quantity` as last_qty 
    -> FROM InventoryTest_history i 
    -> CROSS JOIN (SELECT @last_qty := '', @sold_qty := 0) as init 
    -> ORDER BY `timestamp`; 
+----+-------------+----------+---------------------+----------+----------+ 
| id | SKU   | Quantity | timestamp   | sold_qty | last_qty | 
+----+-------------+----------+---------------------+----------+----------+ 
| 1 | (L) U-Joint |  985 | 2016-12-27 10:08:58 |  0 |  985 | 
| 2 | (L) U-Joint |  960 | 2016-12-28 10:09:52 |  25 |  960 | 
| 3 | (L) U-Joint |  955 | 2016-12-29 16:01:02 |  5 |  955 | 
+----+-------------+----------+---------------------+----------+----------+ 
3 rows in set (0,00 sec) 

mysql> 

Answer 2

一個選項將使用相關的子查詢。對於大集合來說，這當然不是最有效的方法，但對於返回的有限數量的行來說，這是可行的。

似乎有一個序列需要處理和返回行，但沒有ORDER BY子句。從示例數據看來，它看起來像是timestamp列升序，Quantity列降序和/或id升序。我們只是猜測。

假設序列是timestamp和id，並假設(timestamp,id)元組是唯一的......這些都是相當合理的假設，但他們都只是假設，以及相當大的...

SELECT ih.Quantity 
    , ih.timestamp 

    , IFNULL(
       (SELECT pr.Quantity 
        FROM InventoryTest_history pr 
        WHERE pr.SKU = ih.SKU 
         AND pr.timestamp <= ih.timestamp 
         AND (pr.timestamp < ih.timestamp OR pr.id < ih.id) 
        ORDER BY pr.SKU DESC, pr.timestamp DESC, pr.id DESC 
        LIMIT 1 
       ) 
       - ih.Quantity 
     ,0) AS sold_qty 

    FROM InventoryTest_history ih 
WHERE ih.SKU = '(L) U-Joint' 
ORDER BY ... 

---

對於以足夠的性能最好的射手，我們希望可用於相關子查詢覆蓋索引，作爲一個合適的索引的例子：

... ON InventoryTest_history (SKU, timestamp, id, quantity) 

如果應將行視爲按(timestamp,id)以外的順序排序，則需要修改ORDER BY和子查詢中WHERE子句中的條件。

---

這只是一種可能的方法。還有其他查詢模式會返回相同的結果。